Hi, try to write here: https://campaign.optum.com/optum-outcomes/what-we-do/health-surveys/sf-36v2-health-survey.html

They help me a lot with the evaluation.

PCS is calculated from PF, RP, BP and GH and MCS is calculated from VT, SF, RE and MH

Hi Vidyasgar,

did you get an answer to your question.. i'm facing the same queries. i have a project where i use the MSC and i'm not sure how to create the total score from the four sub dimensions. please if you find any helpful resources share it with me .. thanks

Hi Vidyasgar,

Did you find the answer? I have the same question.

Thanks!

the sf-36 has two versions the first version was called rand sf-36 to which there is no pcs or mcs and these scores calculated only for sf-36 version 2

The eight SF-36 domains measure different faces of the quality of life and had been clustered into two main groups: Physical and mental health. If you use the PCA with the 8 domain scores, most of the time you will obtain two main components (try the Varimax rotation). These domains "live" in specific math spaces with more or less reproducible distances and it is easy to differentiate the two main clusters (physical and mental in two PCA components). However, in daily life these domains are not uncorrelated, therefore, I prefer to use Oblimin rotation (delta=0) and work any secondary analysis with the obtained scores (you should not feel uncomfortable with some overlap between the domains).

If you understand STATA this should help:

/*Should we also derive the two summary scores, PCS and MCS?

They are somewhat controversial

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5029523/

Here is the method used by Mobilize Boston:

It uses the orthogonal rotation only

*/

*** 1) TRANSFORM SUBSCALE SCORES TO Z-SCORES *******

*** US GENERAL POPULATION MEANS AND SD'S ARE USED HERE *******

*** (NOT AGE/GENDER BASED) ***********************************

*** SUBTRACT US POPULATION MEAN AND DIVIDE BY SD

gen PF_Z = (sf_36_sub_phys_func - 81.18122) / 29.10588

gen RP_Z = (sf_36_sub_phys_role_lim - 80.52856) / 27.13526

gen BP_Z = (sf_36_sub_pain - 81.74015) / 24.53019

gen GH_Z = (sf_36_sub_gen_health - 72.19795) / 23.19041

gen VT_Z = (sf_36_sub_energy - 55.59090) / 24.84380

gen SF_Z = (sf_36_sub_social - 83.73973) / 24.75775

gen RE_Z = (sf_36_sub_emot_role_lim - 86.41051) / 22.35543

gen MH_Z = (sf_36_sub_emot_well - 70.18217) / 20.50597

*** 2) CREATE PHYSICAL AND MENTAL HEALTH COMPOSITE SCORES: **********

*** MULTIPLY Z-SCORES BY VARIMAX-ROTATED FACTOR SCORING **********

*** COEFFICIENTS AND SUM THE PRODUCTS ****************************

*** THESE US FACTOR SCORES ARE A LITTLE DIFFERENT FROM UK SCORES *

*** SEE Jenkinson_1999 J PUB HEALTH MED 21(4): 372-376 *********

#delimit ;

gen sf_36_PCS = (PF_Z*0.42402) +

(RP_Z*0.35119) +

(BP_Z*0.31754) +

(GH_Z*0.24954) +

(VT_Z*0.02877) +

(SF_Z*-.00753) +

(RE_Z*-.19206) +

(MH_Z*-.22069) ;

gen sf_36_MCS = (PF_Z*-.22999) +

(RP_Z*-.12329) +

(BP_Z*-.09731) +

(GH_Z*-.01571) +

(VT_Z*0.23534) +

(SF_Z*0.26876) +

(RE_Z*0.43407) +

(MH_Z*0.48581) ;

#delimit cr

*** 3) TRANSFORM COMPOSITE SCORES TO T-SCORES: ******

replace sf_36_PCS = 50 + (sf_36_PCS * 10)

replace sf_36_MCS = 50 + (sf_36_MCS * 10)

label variable sf_36_PCS "SF36 PHYSICAL COMPONENT SCORE(T-SCORE)"

label variable sf_36_MCS "SF36 MENTAL COMPONENT SCORE (T-SCORE)"

sum sf_36_PCS sf_36_MCS

/*

OUR VALUES:

Variable | Obs Mean Std. Dev. Min Max

-------------+---------------------------------------------------------

sf_36_PCS | 372 46.67904 9.925978 16.46143 59.64453

sf_36_MCS | 372 55.20761 8.39028 21.94317 70.35992

Compare to Mobilize Boston:

Variable | Obs Mean Std. Dev. Min Max -------------+--------------------------------------------------------

vdPCS | 1037 43.64571 10.30067 12.66702 63.38371

vdMCS | 1037 55.15717 8.343987 16.08196 74.91134

*/

Hi Cota! Did you find the response to your questions? It would be great if you can share.

I have the same query. Can you share the answer please

If you are using R, STATA or SPSS, there are already code implemented to calculate this....

Hello,

Did you end up finding an answer to this?

I am using the RAND Short form Health survey (version 1.0) and was wondering if I can calculate the Mental and physical component summary scores? Or can you only calculate the 8 different subscales?

Scoring of the two composited MOS SF-36 subscales involves three steps: 1) standardization of the scales (z-scores), 2) aggregation of scale scores, and 3) transformation of summary scores (T-scores) (Ware et al., 1994). For each domain (PCS and MCS), the mean score is 50 with a standard deviation of 10. Thus, higher scores indicate better QOL (score range 0-100).

Ware, J. E., Kosinski, M., & Keller, S.K. (1994). SF-36 physical and mental health summary scales: A user’s manual. The Health Institute.

Ware J. E. (2000). SF-36 health survey update. Spine, 25(24), 3130–3139. https://doi.org/10.1097/00007632-200012150-00008

Hope this works. Best,

Thank you for your response.

Does this work if I have used the RAND SF36 version?

Hi,

I have the same question about RAND sf-36 and possibility to calculate PCS and MCS. Found one article, maybe it would help

https://www.rand.org/content/dam/rand/pubs/reprints/2008/RAND_RP1309.pdf

Not sure, but my rationalize for this is:

The RAND SF-36 is the same instrument as the SF-36v1 (or MOS SF-36). It was created in 1990. The SF-36v2 was created in 1996. The reference manual was developed in 1994, therefore, the scoring for the MCS and PCS are for version 1 because version 2 was not developed yet.

Ware, J. E., Kosinski, M., & Keller, S.K. (1994). SF-36 physical and mental health summary scales: A user’s manual. The Health Institute.

Please let me know if anybody has another opinion.

Thank you for responses:)

In RAND official webpage in scoring rules for the RAND 36-item health survey (Version 1.0) mentioned: "we recommend that responses be scored as described below. A somewhat different scoring procedure for the MOS SF-36 has been distributed by the IRCHCA. Because the scoring method described here (a simpler and more straightforward procedure) differs from that of the MOS SF-36, persons using this scoring method should refer to the instrument as RAND SF-36. "

So, I assume there is a difference but what exactly differ - not sure.

https://www.rand.org/health-care/surveys_tools/mos/36-item-short-form/scoring.html

Hi all,

Sorry, I forgot to provide the tools for this analysis.

Here is a comprehensive code for SF-36 analysis in R: https://github.com/lbraglia/lbscorer/blob/master/R/sf36.R.

For SAS: https://www.rand.org/content/dam/rand/www/external/health/surveys_tools/kdqol/kdqol13.sas

Thank you. I hope this help.

Best,

Dear all,

I am not sure whether this question raised has been solved, but I also have the same problem. MOS SF-36 and Rand SF-36 have identical thirty-six questions but with different scoring methods. I found this paper (attached), and maybe it will explain why there is the difference and what I think is in which scoring method you choose to calculate the scores for the 8 dimension and the 2 component summarize.

However, I still don't get the idea who to calculate the summary scores (PCS and MCS) by using RAND SF-36 scoring method, which in the article it was written (on page 221 last paragraph and page 223 left upper paragraph):

Results of previous factor analyses suggested a procedure that might be useful for deriving physical and mental health composites. The eight multi-item short-form scales were rotated obliquely (Promax rotation) to obtain physical health and mental health factor scores and these scores were transformed linearly to T-scores.

Would anyone be kind enough to instruct me how to do this accordingly? Thank you.

Elizabeth Procter-Gray Dear Professor, I would like to extend my question in response to your reply to this query. I found in your response you recommend calculating a t-score for the two summary components (PCS and MCS). And as I know t-score will generate a mean of 50 and different SD at the end, which means the means of PCS and MCS will be the same (t-score). But, in your later written sentences you compare your valuanother studyes with other study by using PCS and MCS scores, supposed this should be the t-scores, right? but why is it with different means? could you please kindly instruct me? thank you.

Hello

Did anyone solve how to score the MCS and PCS using the RAND SF36 version 1.0? I cannot seem to find instructions on how to calculate the summary scores.